Cutting Edge: Glycolytic Metabolism and Mitochondrial Metabolism Are Uncoupled in Antigen-Activated CD8⁺ Recent Thymic Emigrants

T cells were once thought to be fully mature following thymic selection and egress (1). This notion has been progressively revised as improved models have allowed for the study of intrinsic peripheral T cell age (reviewed in Ref. 2). The Rag2p-GFP transgenic (Tg) mouse (3) has been used to isolate live, untouched recent thymic emigrants (RTEs), facilitating previously unfeasible experiments (4). Using this model, it has become clear that for the first 3 wk in the lymphoid periphery, T cells occupy a distinct phenotypic and functional niche. Relative to mature T cells, CD8⁺ RTEs are poor proliferators and producers of effector cytokines (IFN-γ, TNF-α, and IL-2) following activation in the absence of inflammation (4–7). Critically, Ag exposure during the period when T cells occupy the RTE compartment imparts lasting functional differences that distinguish RTE-derived memory cells from mature T cell–derived memory cells (8). Importantly, the unique physiology of RTEs is not just a feature of the mouse but also of humans (9–12).

RTE maturation is independent of many of the signals required for mature but naive (MN) T cell homeostasis including IL-7, self-MHC molecules, and CD80/CD86 (13, 14). However, thymic egress and access to secondary lymphoid organs are strictly required for RTE maturation (15). Recently, the lymphocyte migration factor sphingosine 1-phosphate, which is present at high concentrations in the blood, has been shown to regulate MN T cell homeostasis by maintaining mitochondrial content (16). These data underscore the possibility that RTE maturation has a metabolic component.

MN T cells undergo profound metabolic reprogramming following activation (17) that is associated with an increase in the rate of aerobic glycolysis (18). However, even in the presence of this large glycolytic flux, mitochondrial metabolism remains critically important both for T cell activation and effector function (19, 20). The requirement for ongoing respiration in activated T cells extends beyond the requisite ATP generation (reviewed in Ref. 21). In CD4⁺ T cells, production of mitochondrial reactive oxygen species is required for NFAT-mediated IL-2 production (20). Additionally, mitochondria-derived acetyl-CoA is critical for histone acetylation at the ifng locus in the context of ongoing aerobic glycolysis (22). Mitochondrial metabolism also regulates memory T cell formation (23–25). Specifically, mitochondrial morphology distinguishes CD8⁺ effector from memory T cells. The former have round mitochondria with loose cristae and the latter have elongated mitochondria with tight cristae, in which key membrane components are brought into close spatial proximity, thereby enhancing the efficacy of electron transport (26–28). Thus, mitochondria represent a central hub through which T cell fate decisions are orchestrated.

In addition to signals triggered through the TCR, costimulatory signals delivered through CD28 play a well-established role in regulating the induction of glycolytic metabolism in activated T cells (29) and an emerging role in modulating mitochondrial function (30). CD28 signals are required for the establishment of tight cristae and carnitine palmitoyltransferase 1a (Cpt1a)–mediated spare respiratory capacity (SRC). Additionally, 4-1BB signaling has been implicated in the regulation of mitochondrial metabolism (31), being critical for optimal oxidative phosphorylation (OxPhos) via peroxisome proliferator–activator receptor γ coactivator-1α (PGC-1α)–mediated accumulation of mitochondrial mass.

We recently found that activated CD8⁺ RTEs have a reduced rate of glycolysis relative to mature T cells, a defect that underlies their diminished IFN-γ production (32). Activation in the presence of exogenous IL-2 corrects this deficit. As mitochondria are central drivers of effector function, we now assess mitochondrial metabolism in RTEs to better understand their unique biology. In this study, we show that activated CD8⁺ RTEs have reduced OxPhos and SRC relative to their more mature counterparts. RTEs fail to increase glutaminase (GLS) expression and mitochondrial mass following activation, defects that persist even in the presence of exogenous IL-2. These alterations are driven by reduced CD28 signaling in RTEs. Collectively, these results suggest that although glycolytic metabolism in activated CD8⁺ RTEs is subject to modulation by IL-2, mitochondrial metabolism appears to be “hard wired” differently in RTEs than in mature T cells. These results have profound implications for T cell responses that are dominated by RTEs, such as those in neonates and in adults recovering from therapeutic or accidental lymphodepletion.

C57BL/6J, B6.129P2-Tcrb^tm1MomTcrd^tm1Mom/J (TCRβδ^−/−), B6.129S4-Cd80^tm1ShrCD86^tm2Shr/J (CD80/86^−/−), and OT-I [B6-Tg (TcraTcrb) 1100Mjb/J] mice expressing a transgene-encoded TCR that recognizes the chicken OVA-derived peptide SIINFEKL in the context of H-2K^b were purchased from The Jackson Laboratory and maintained in-house. CD80/86^−/− mice were cohoused with wild type (WT) controls for >14 d prior to experimentation. Rag2p-GFP Tg mice (3) were backcrossed onto the C57BL/6J background in our laboratory for at least 12 generations. Mice of both sexes were used at 8–10 wk of age. All mice were housed under specific pathogen-free conditions and used in accordance with the University of Washington Institutional Animal Care and Use Committee guidelines.

RTE and MN OT-I Tg T cells were isolated from Rag2p-GFP Tg mice (7) and stimulated (32) as previously described. In brief, sort-purified CD45.1 RTEs and MN T cells were plated at 1 × 10⁵ per well + 5 × 10⁵ irradiated CD45.2 TCRβδ^−/− or T cell–depleted CD80/86^−/− splenocytes as APCs in complete RPMI 1640 (Life Technologies) containing 10% heat-inactivated FBS. Cells were stimulated with 10 nM SIINFEKL (Genmed Synthesis) ± 20 U/ml rIL-2 (Cetus) and maintained for 3 d at 37°C in 7% CO₂.

Fc receptors were blocked using anti-16/32 (2.4G2; BD Biosciences), and cells were stained with fluorescent Abs to detect cell surface expression of 4-1BB (17B5), CD4 (RM4-5), CD11b (M1/70), CD25 (PC61.5), CD44 (IM7), CD45.1 (A20), B220 (RA3-6B2), ICOS (7E.17G9) NK1.1 (PK136), and Ter119 (Ly-76), all from eBioscience, in addition to CD8α (53-6.7), CD28 (37.51), and CD62L (MEL-14) from BD Biosciences. Data were acquired using a FACSCanto II (BD Biosciences) and analyzed with FlowJo software (Tree Star).

Surface-stained cells were fixed and permeabilized using the Cytofix/Cytoperm Kit (BD Biosciences) and stained with Abs against GLS (polyclonal), ATPB (3D3), Cpt1a [EPR12740(B)], and voltage-dependent anion channel 1 (VDAC1; 20B12AF2), all from Abcam, and PGC-1α (D-5; Santa Cruz Biotechnology), p-S6 S240/S244 (D68F8), p–NF-κB (931H1) and p-p38 (D3F9), all from Cell Signaling Technology. For IL-2 staining, cells were plated and activated as above for 6 h; GolgiPlug (BD Biosciences) was added after the first hour. Cells were stained with a labeled Ab against IL-2 (JES6-5H4; eBioscience).

Oxygen consumption rate (OCR) measurements (in picomoles per minute) were generated using an XF24 bioanalyzer (Seahorse Bioscience). T cells were plated (1 × 10⁶ per well) into a XF24 microplate coated with Cell-Tak (Sigma-Aldrich). Cells were allowed to equilibrate in nonbuffered medium (Seahorse Bioscience) containing 10 mM glucose for 30 min in a non-CO₂ incubator. Four baseline OCR readings were acquired before sequential injection of 2.5 μM oligomycin A, 500 nM carbonyl cyanide-4-(trifluoromethoxy)phenylhydrazone, and 2 μM rotenone +2 μM antimycin A, all from Sigma-Aldrich. To determine the contribution of glutamine to OCR, basal OCR was measured in cells plated in nonbuffered medium containing 10 mM glucose and glutamine at either 0.2 mM (−Gln) or 2 mM (+Gln).

CD8⁺ RTEs and MN T cells were stimulated as above, and mitochondria and nuclear DNA were visualized using MitoTracker Red CMXRos (Thermo Fisher Scientific) and NucBlue Live ReadyProbes Reagent (Molecular Probes), respectively. Cells were allowed to settle onto poly-d-lysine (Sigma-Aldrich)–coated glass-bottom petri dishes (MatTek) for 30 min before imaging. A z-series with a 0.3-μm step size was collected at 37°C and 7% CO₂ using a Ti-E widefield microscope with a 100× numerical aperture 1.45 oil objective (Nikon), a solid-state light source (SPECTRA X; Lumencor), and a scientific complementary metal oxide semiconductor camera (Zyla 5.5 megapixel). Mitochondrial morphology was scored in a blinded fashion as fragmented: spherical mitochondria <1.5 μm in diameter; intermediate: oblong and <2.5 μm in length; and reticular: elongated and highly connected.

RTEs and mature T cells were activated as above, and 70-nm sections were imaged using a JEM-1230 transmission electron microscope (JEOL USA) in the University of Washington Vision Science Center. Image analysis was performed blinded on 35–40 measurements of maximum crista width per condition across 20 image fields.

The p values were calculated in Prism (GraphPad) using a paired two-tailed Student t test except where indicated. The p values <0.05 were considered significant.

Our previous work indicated that activated RTEs display glycolytic defects that can be corrected by exogenous IL-2 (32). However, mitochondrial metabolism and glycolytic metabolism play nonredundant roles in T cell activation and effector function (19). Indeed, exogenous IL-2 failed to restore defective IL-2 production by activated RTEs (Fig. 1A). As IL-2 production requires mitochondrial metabolism (20), we assessed the rate of OxPhos, which generates ATP and is coupled to NAD⁺/NADH homeostasis and the generation of mitochondrial reactive oxygen species in activated T cells (21). We found that RTEs and MN T cells measured directly ex vivo displayed equivalent OCRs. However, the robust, activation-dependent induction of OCR seen in mature T cells was significantly attenuated in RTEs (Fig. 1B, left). Additionally, activated RTEs displayed less SRC (a measure of mitochondrial reserve). Importantly, exogenous IL-2 had no effect on RTE basal OCR or SRC (Fig. 1B) despite restoring maximum glycolytic rate (32). The TCA cycle is critical for generating NADH and FADH₂. In activated CD8⁺ T cells, glutaminolysis and fatty acid oxidation are critical for replenishing TCA-cycle intermediates depleted by cataplerotic processes (33). As RTEs display reduced OCR and SRC, we next assessed the expression of rate-determining enzymes for glutaminolysis (GLS) and fatty acid oxidation (Cpt1a). To this end, we assessed enzyme expression in RTEs and mature T cells left resting or activated ± exogenous IL-2. Although resting RTEs and mature T cells express similar levels of GLS (Fig. 1C) and Cpt1a (Supplemental Fig. 1A), enzyme induction is significantly diminished in activated RTEs. Critically, IL-2 fails to restore GLS expression in RTEs (Fig. 1C). To confirm that altered glutamine flux drives the differences in OCR seen in RTEs and mature T cells, OCR was measured in cells activated in medium containing glutamine and then plated in the presence of low (0.2 mM; −Gln) or high (2 mM; +Gln) glutamine. Activated mature T cells had a significantly reduced OCR in the presence of low glutamine (Fig. 1D). By contrast, glutamine access did not impact the OCR of activated RTEs. These results suggest that the defective OCR observed in RTEs primed in the presence of IL-2 is associated with a reduced rate of glutaminolysis. Interestingly, GLS is a Myc target gene (17) and IL-2 promotes Myc expression (34) and restores Myc protein levels in RTEs (32). This discrepancy suggests that the GLS locus may be in an altered epigenetic state in RTEs, as has been observed for certain cytokine loci (35). Alternatively, Myc may regulate a subtly different set of genes in RTEs and mature T cells.

Respiratory capacity in activated CD8⁺ T cells is regulated by mitochondrial content (23) and morphology (26, 30). To determine mitochondrial mass, RTEs and mature T cells were left resting or activated ± exogenous IL-2 and the level of VDAC1 was determined. Resting CD8⁺ RTEs and mature T cells had equivalent VDAC1 levels, similar to recent observations in resting CD4⁺ RTEs (36). However, upon activation, RTEs expressed significantly less VDAC1 than their mature T cell counterparts (Fig. 2A). This defect was also reflected in the expression of the ATPB subunit of Complex V (Fig. 2B), Tom20, and staining with the mitochondrial membrane potential-sensitive dye MitoTracker Red (data not shown). Critically, activation of RTEs in the presence of IL-2 failed to correct this defect in mitochondrial mass (Fig. 2A, 2B). By MitoTracker staining and live-cell microscopy, the majority of activated mature T cells contained mitochondria that were fragmented, with few cells having mitochondria of intermediate length or reticular appearance (Fig. 2C). In contrast, fewer activated RTEs had fragmented mitochondria, and a large proportion contained mitochondria of intermediate length or reticular appearance. Crista structure is generally coupled to mitochondrial morphology (26), and both resting RTEs and mature T cells had tight cristae (Fig. 2D). Upon activation, mature T cells displayed greater crista width than RTEs, and this difference was unaffected by activation in the presence of IL-2. Thus, activated RTEs have more elongated mitochondria with tighter crista structure (parameters associated with greater respiratory capacity); however, this apparent respiratory advantage is offset by a profound reduction in mitochondrial mass.

Reduced mitochondrial mass even in the presence of IL-2 implies a defect in mitochondrial biogenesis following activation, a process that is regulated by PGC-1α (37–39). Resting RTEs and mature T cells had equivalent levels of PGC-1α; however, activated RTEs expressed significantly less PGC-1α than their more mature counterparts (Fig. 2E). Importantly, the addition of IL-2 failed to restore PGC-1α levels in RTEs, suggesting that the reduced mitochondrial mass in RTEs is the result of lower PGC-1α expression.

Seminal work has shown that the costimulatory receptor CD28 regulates glycolytic metabolism in CD8⁺ T cells (29), a role that has been expanded to include respiratory metabolism (30, 31). To determine if costimulatory receptors contribute to the distinct mitochondrial metabolism in RTEs, we assessed expression levels of 4-1BB, ICOS, and CD28. Compared with mature T cells, resting RTEs expressed equivalent levels of 4-1BB (Fig. 3A) and ICOS (Fig. 3B) but had a subtle, reproducible reduction in CD28 expression (Fig. 3C) (4). Upon activation, RTEs expressed significantly less 4-1BB, ICOS, and CD28 than their more mature T cell counterparts. However, in the presence of IL-2 (which fails to restore OCR and SRC), RTEs expressed less 4-1BB but equivalent ICOS and CD28; hence, reduced CD28 expression on naive cells and reduced 4-1BB expression on activated cells correlates with the diminished respiratory capacity in activated RTEs.

Compared with mature T cells, resting RTEs express less CD28 and have more elongated mitochondria with tighter crista structure, alterations seen in CD8⁺ T cells primed without costimulation (30). To determine if altered CD28 signaling accounts for the diminished mitochondrial mass in RTEs, cells were left resting or activated with Ag presented by WT or CD80/86^−/− APCs. As expected, RTEs have lower mitochondrial mass than mature T cells when activated with WT APCs. By contrast, RTEs and mature T cells had equivalent mitochondrial mass when activated with CD80/86^−/− APCs (Fig. 4A). Mitochondrial morphology in both activated RTEs and mature T cells was unaffected by the loss of CD28 signaling at this timepoint (Figs. 2C, 4B). By contrast, crista width remained tight in both RTEs and mature T cells activated with CD80/86^−/− APCs (Fig. 4C). In support of these findings, RTEs and mature T cells expressed equivalent PGC-1α levels when activated with CD80/86^−/− APCs (Fig. 4D). These data suggest that alterations in CD28 signaling account for the diminished mitochondrial mass observed in activated RTEs. To determine whether the reduced CD28 expression by RTEs translates to quantitatively different signaling, we assessed the level of p-S6 as a readout of CD28 signaling intensity. As observed previously (32), RTEs activated with WT APCs have reduced p-S6 levels compared with their mature T cell counterparts (Fig. 4E). However, RTEs and mature T cells had similarly reduced levels of p-S6 when activated with CD80/86^−/− APCs. These data argue that the reduced surface expression of CD28 on RTEs results in a quantitatively reduced downstream signal. Additionally, 4-1BB can support mitochondrial biogenesis in activated T cells (31). However, 4-1BB transduces a similar downstream signal in RTEs and mature T cells as measured by p-p38 (Supplemental Fig. 1B) and p–NF-κB (Supplemental Fig. 1C). Blocking 4-1BBL/4-1BB interactions did not affect VDAC1 levels in either RTEs or mature T cells (Supplemental Fig. 1D), further suggesting that diminished signaling via CD28, but not 4-1BB, accounts for the reduced mitochondrial mass observed in RTEs.

In sum, our data show that activated RTEs have defective OxPhos even in the presence of exogenous IL-2. This reduced respiratory rate is due to defective GLS induction and reduced accumulation of mitochondrial mass following activation. Altered CD28 signals account for the reduced PGC-1α expression and diminished accumulation of mitochondrial mass in RTEs. As shown schematically in Supplemental Fig. 2, our study highlights that although defective glycolytic metabolism in RTEs is correctable by the addition of IL-2, the altered mitochondrial metabolism in these transitional T cells appears to be hard wired. The reduced mitochondrial content in activated RTEs may also confer RTEs with a survival advantage, as mitochondria serve as central hubs in the regulation of apoptosis (40). Last, this work has important implications in the context of antitumor immune responses in which mitochondrial metabolism in CD8⁺ T cells appears to be the central driver of disease outcome (31).

We thank Edward Parker, Daciana Margineantu, and David Hockenbery for technical assistance and Ramon Klein-Geltink and Erika Pearce for helpful discussions.

The authors have no financial conflicts of interest.